Modern OS, Legacy Iron: Why I’m Betting on AlmaLinux 9 in an Aging HPE Data Center

The Modernization Dilemma

As enterprise infrastructure continues to stretch beyond traditional lifecycle timelines, we’re all facing the same dilemma: how do we modernize without replacing everything? It’s easy to talk about moving fast, upgrading operating systems, or deploying open-source solutions. But the reality on the ground is different.

At the heart of that reality in my environment are hundreds of Gen9 and Gen10 HPE servers running reliably, but backed by hardware that’s aging out of support. When CentOS reached its end-of-life and AlmaLinux 9 rose as a viable, RHEL-compatible alternative, I saw an opportunity to evolve. But upgrading isn’t just a software story—it’s a hardware gauntlet.

The Compatibility Gauntlet: OS vs Hardware

AlmaLinux 9 is stable, powerful, and community-backed. It should be a clear win. But here’s the rub: when you install AlmaLinux 9 on Gen9 or early Gen10 HPE servers with Broadcom 7300-series NICs, you’re met with driver mismatches, firmware incompatibilities, and a system that doesn’t always recognize its own interfaces.

These NICs, especially the 7322 models, rely on the bnxt_en driver—which in kernel 5.14+ (used by AlmaLinux 9) has evolved past the point of supporting older firmware revisions. We’re forced to either flash firmware, rip and replace NICs, or draft entirely new system profiles just to get the OS online and functional.

The Cost of Staying Put

Our team has resorted to replacing aging NICs with newer models like the x2522-25G+ or x2541 series, which introduces its own pain:

Server rebuilds

BIOS tuning

Software-level flashing

And this isn’t a one-time fix. It’s a repeated cycle across dozens of production nodes. We are hard-pressed to take down working servers for the sake of an OS upgrade—but staying on unsupported platforms like RHEL/CentOS 7 isn’t an option either.

In some cases, the only path forward is to decommission fully functioning Gen9 servers and replace them with Gen11 or Gen12 platforms—a move that contradicts our original goal of minimizing CapEx and extending the value of existing infrastructure.

Strategic Modernization vs Forced Replatforming

This is the crossroads many data center leaders find themselves in today. We aim to modernize without reinventing. Yet when the foundational hardware starts resisting, you’re not upgrading anymore—you’re replatforming.

My message to others in the same situation: plan proactively. Know your hardware’s limits before committing to an OS refresh. Build a compatibility matrix. Test firmware and NIC behavior in staging. And, most importantly, have a plan for when modernization turns into strategic replacement.

AlmaLinux 9 Is Still the Right Move—With Conditions

Despite these challenges, AlmaLinux 9 remains a strategic asset. For newer hardware, it’s an ideal drop-in replacement. For edge workloads, dev clusters, or hybrid deployments, it’s flexible and cost-effective. But in legacy environments, it’s not plug-and-play.

For it to succeed, you need to:

Update system firmware (use the latest HPE SPP)

Replace or reflash unsupported NICs

Rebuild system profiles and validate network performance

Prepare your team for more than just a yum upgrade

Closing Thoughts: Upgrades Aren’t Just Technical—They’re Strategic

This process has taught me that infrastructure upgrades are never just about technology. They’re about timing, resourcing, and making tradeoffs that preserve business continuity while positioning for growth.

AlmaLinux 9 represents a path forward. But it’s not a shortcut. And in a world where everything from firmware to drivers is a moving target, the real upgrade is our mindset.

1 Like

Well said.

However, are the old NICs such that ELRepo has not build them a driver?
ELRepo does build kernel modules for devices (old and new) that RH does not support in RHEL.

1 Like

Thanks, appreciate that.

You’re absolutely right — ELRepo has done a great job over the years filling in the gaps where RHEL (and by extension Alma) no longer maintains support. In fact, our team did explore ELRepo’s kmod-bnxt_en as a workaround for the 7300-series NICs.

In some edge cases, it got us up and running temporarily, but it wasn’t a silver bullet. We still ran into firmware mismatches and limitations with advanced features like SR-IOV and RDMA, especially on Gen9 hardware. It became clear that the broader system profile needed to evolve, not just the driver.

So while ELRepo is absolutely a solid stopgap for many use cases, in our case it wasn’t enough to avoid the inevitable shift to newer NICs and eventually Gen11+ systems.

Appreciate you raising that point — it’s definitely part of the broader toolkit for environments trying to squeeze more life out of legacy hardware.

2 Likes

So, the NIC’s issue is that you have been neglecting to keep the firmware up to date. Good luck with your support call to HPE. You can expect to hear, “It appears you are not on the latest firmware, please upgrade ” during the call. I bet your iLOM and BIOS firmwares are all doggy ancient too. Your approach to firmware upgrades is completely unprofessional in the cybersecurity landscape we find ourselves in 2025.